EP3734822A1

EP3734822A1 - Methods and apparatus for active flux control of an output transformer connected to a power converter

Info

Publication number: EP3734822A1
Application number: EP19171648.9A
Authority: EP
Inventors: Julian Eduardo Bonilla Alarcon
Original assignee: GE Energy Power Conversion Technology Ltd
Current assignee: GE Energy Power Conversion Technology Ltd
Priority date: 2019-04-29
Filing date: 2019-04-29
Publication date: 2020-11-04
Anticipated expiration: 2039-04-29
Also published as: WO2020221828A1; EP3734822B1

Abstract

An active flux control method is described. The active flux control method injects a DC current component into an output transformer (10) that is connected to a power converter in order to reduce the saturation of the output transformer during a network fault. The method includes using a flux controller (20) to derive an estimated magnetisation curve of the output transformer using a closed-loop estimator comprising a flux model (21) and a modelled magnetisation flux (22). A DC component of the estimated magnetisation flux is calculated. Current control is carried out using a current controller (30) that controls the power converter to inject a DC current component into the output transformer. The DC current component is calculated from the DC component of the estimated magnetisation flux.

Description

Technical Field

The present invention relates to active flux control of an output transformer connected to a power converter (e.g., a static frequency converter) in fault situations. The present invention is particularly useful in conventional single-phase alternating current (AC) electric rail networks.

Background Art

In a conventional electric rail network, the overhead line (sometimes called a "catenary") carries a single-phase alternating current (AC) rail network voltage with a fundamental frequency (e.g., 16.7, 25 or 50 Hz). The overline line is connected to a series of substations. Each substation may include a power converter with a direct current (DC) link that includes one or more capacitors. In particular, the overhead line may be connected to the AC terminals of a DC/AC output converter by means of an output transformer. The DC terminals of the output converter may be connected to the DC terminals of an AC/DC input converter by means of the DC link. The AC terminals of the input converter may be connected to a three-phase AC power network or utility grid operating at a fundamental frequency (e.g., 50 or 60 Hz) by means of an input transformer. The single-phase rail network voltage may therefore be derived from the three-phase grid voltage by means of the power converter.
The present invention aims to reduce the saturation of the output transformer during a fault in the electric rail network, e.g., in the overhead line, by actively controlling the flux in the output transformer.
In particular, the rail network voltage may sometimes experience a voltage dip either because of a network fault or during testing of protection systems for the electric rail network. Abrupt voltage disruption may saturate the output transformer and other inductive elements in the electric rail network, which may lead in turn to non-linear behaviour of the output transformer. Non-linear behaviour affects the network resonance frequencies such that, during the fault, oscillations at non-fundamental frequencies appear in the network voltage and network current. This may significantly affect the proper operation of the protection systems for the electric rail network. Protection systems typically use voltage and current measurements to determine the exact location of the fault so that the faulty network section may be isolated. But it may be difficult for the protection systems to make the correct determination unless the voltage and current measurements have sinusoidal waveforms at the fundamental frequency. Distorted waveforms caused by the non-linear behaviour of the output transformer may therefore lead to the incorrect isolation of network sections and to delays in restoring power to the electric rail network after the fault.
In light of the above, there is a need for an improved method of active flux control suitable for electric rail networks and an apparatus for implementing any such method.

Summary of the invention

The present invention provides a method of active flux control for an output transformer connected to a power converter (e.g., a static frequency converter) comprising the steps of:

deriving an estimated magnetisation flux of the output transformer using a closed-loop estimator comprising a flux model and a modelled magnetisation flux;
calculating a direct current (DC) component of the estimated magnetisation flux; and
carrying out current control by injecting a DC current component into the output transformer, wherein the DC current component is calculated from the DC component of the estimated magnetisation flux.

The method of the present invention is advantageous in that it may provide improved flux control by controlling the power converter during network faults to inject a DC current component into the output transformer. In particular, the method may quickly reduce the saturation of the output transformer during a network fault. The method of the present invention is of particular use with an output transformer connected to a static frequency converter of an electric rail network that is supplied from a three-phase alternating current (AC) supply with a first fundamental frequency and where the electric rail network operates with a single-phase AC with a second fundamental frequency.
The method may include the step of obtaining a measure of one or more of a primary current, a secondary current, a filter current, and a converter voltage of the output transformer, which can be used by the closed-loop estimator to derive the estimated magnetisation flux of the output transformer. The values of the primary current, the secondary current, the filter current, and the converter voltage of the output transformer may be obtained in any manner apparent to the skilled person. The values may be measured directly or indirectly or may be estimated using known methods. The primary current is the current in the primary coil of the output transformer. The secondary current is the current in the secondary coil of the output transformer. The filter current and converter voltage may be understood with reference to Figure 2, which shows a simplified schematic of an output transformer of a static frequency converter and is discussed below.
The estimated magnetisation flux of the output transformer is derived by a closed-loop estimator that comprises both the flux model and a modelled magnetisation curve that can be used to derive the modelled magnetisation flux. A closed-loop estimator is advantageous over an open-loop estimator comprising only a flux model as it has a reduced steady state error.
The flux model of a closed-loop estimator may comprise any suitable dynamic model that provides an estimate of the magnetisation flux of the output transformer. For example, the flux model may comprise an open-loop estimator that will be known to the skilled person with an additional input derived from the difference between an output of the flux model and the modelled magnetisation flux. The flux model inputs may be the measured secondary current i_s, the measured converter voltage u_s, and a feedback input ΔΦ_m derived from the difference between an output of the flux model (i.e., the estimated magnetisation flux Φ̂_m) and the modelled magnetisation flux Φ_m.
If the secondary current i_s and converter voltage u_s are known, an open-loop estimator may estimate the magnetisation flux Φ̂_m of the output transformer from the following equation: $u_{s} = L_{s} \frac{{di}_{s}}{dt} + R_{s} i_{s} + \frac{d {\hat{Φ}}_{m}}{dt}$
where L_s is the secondary inductance of the output transformer and R_s is the secondary resistance of the output transformer.
As u_s and i_s are both known, the observer state is defined as: $x = L_{s} i_{s} + {\hat{Φ}}_{m}$
The observer state equation and output equations are: $\frac{dx}{dt} = u_{s} - R_{s} i_{s}$
${\hat{Φ}}_{m} = x - L_{s} i_{s}$
The modelled magnetisation flux Φ_m provides an output that is an open-loop estimate of the magnetisation flux. This output may be obtained in any manner apparent to the skilled person. For example, in one arrangement, the modelled magnetisation flux Φ_m may be obtained from a magnetisation curve specifically adapted to the output transformer and an estimate of a magnetisation current of the output transformer.
The magnetisation current i_m of the output transformer may be estimated from the obtained values of the primary current i_p, secondary current i_s, and filter current i_f of the output transformer as follows: $i_{m} = i_{s} - i_{p} - i_{f}$
The magnetisation curve of the output transformer may be obtained from measurements or derived from a suitable model. The magnetisation curve provides a function f_sat for estimating the magnetisation flux Φ_m from the estimated magnetisation current i_m: $Φ_{m} = f_{sat} (i_{m})$
As will be readily understood by the skilled person, there are several well-known methods by which the DC component of the estimated magnetisation flux Φ̂_m may be estimated. All such methods are inherently suitable for use with the method of the present invention. In one arrangement, the DC component of the estimated magnetisation flux (i.e., Φ̂_dc) is estimated using a moving average filter (or MAF) over N samples of the form: ${\hat{Φ}}_{dc} (t) = \frac{1}{T} \int_{0}^{T} {\hat{Φ}}_{m} (t - τ) dτ$
This equation is cast in a discrete-time recursive form for implementation in the method of the present invention.
Alternative methods by which the DC component of the estimated magnetisation flux may be estimated include, but are not limited to, a low pass filter wherein the cut-off frequency of the filter is much smaller than the frequency of the rail network voltage or a notch filter at the frequency of the rail network voltage wherein the output of the notch filter is subtracted from the total estimated magnetisation flux, for example.
In order to avoid constant operation of the active flux control method, and the resulting constant injection of a DC current component into the output transformer, it may be preferable that the method further comprises an activation function that activates the injection the DC current component into the output transformer when the DC component of the estimated magnetisation flux rises above a first threshold value, and that subsequently deactivates the flux controller when the DC component of the estimated magnetisation flux drops below a second threshold value. The second threshold value is lower than the first threshold value. In other words, the method may comprise an activation function that controls the method so that whilst the DC component of the estimated magnetisation flux remains below a first limit value no DC current component is injected into the output transformer. When the DC component of the estimated magnetisation flux rises above the first threshold value, the method of the present invention may activate the flux controller to inject a DC current component into the output transformer. The injection of the DC current compound into the output transformer will preferably continue until the DC component of the estimated magnetisation flux drops below the second threshold value as a result of the active flux control. After the DC component of the estimated magnetisation flux drops below the second threshold value the flux controller may be deactivated and is not reactivated unless and until the DC component of the estimated magnetisation flux rises back above the first threshold value.
An activation function may be of the form: $Φ_{{CTRL}_{ON}} = {\begin{matrix} 1, if {\hat{Φ}}_{dc} > x_{1} \\ 0, if {\hat{Φ}}_{dc} \leq x_{2} \end{matrix}$
where x₁ is the first threshold value and x₂ is the second threshold value, with x₁ greater than x₂. The first threshold value and the second threshold value may be set at any appropriate values. The first threshold value may be between 20% and 90% saturation of the output transformer. The second threshold value may be between 10% and 80% of the saturation of the output transformer. In order to ensure good operation of the active flux control method, it is preferable that the second threshold value is less than the first threshold value by at least 5% of the saturation of the output transformer.
In one arrangement, the first threshold value is 50% of the saturation of the output transformer and the second threshold value is 40% of the saturation of the output transformer so that: $Φ_{{CTRL}_{ON}} = {\begin{matrix} 1, if {\hat{Φ}}_{dc} > 0.5 p . u . \\ 0, if {\hat{Φ}}_{dc} \leq 0.4 p . u . \end{matrix}$
In an alternative arrangement, the first threshold value is 60% of the saturation of the output transformer and the second threshold value is 50% of the saturation of the output transformer so that: $Φ_{{CTRL}_{ON}} = {\begin{matrix} 1, if {\hat{Φ}}_{dc} > 0.6 p . u . \\ 0, if {\hat{Φ}}_{dc} \leq 0.5 p . u . \end{matrix}$
The current control may be carried out by:

calculating a corrective current from the DC component of the estimated magnetisation flux (e.g., that represents a DC current component to be injected into the output transformer);
providing a reference current (e.g., that represents a sinusoidal current component to be injected into the output transformer); and
calculating the total current to be injected into the output transformer from the corrective current and the reference current.

In order to provide the most appropriate current injection into the output transformer it may be advantageous that the method further comprises the step of using a tunable controller to control the magnitude and/or duration of the DC current component that is injected into the output transformer. The tunable controller may act on the DC component of the estimated magnetisation flux and the output of the tunable controller may be the corrective current.
If the DC component of the estimated magnetisation flux is estimated using a MAF, the tunable controller may be a simple proportional control. This is possible as in a MAF there is an integral action from the converter voltage to the estimated magnetisation flux. A tunable controller that is a simple proportional controller may be represented by: ${Δi}_{s} = - K_{Φ}^{*} {\hat{Φ}}_{dc}$
where Δi_s is the corrective current and $K_{Φ}^{*}$
is the gain of the tunable controller.
This is possible because the MAF is integrating the flux over a time period T and, as a result, there is an integral part for the control provided by the MAF. Further, there is no derivative action from the converter voltage to the output transformer magnetisation flux. This allows a simple proportional control, of the type set out above, to guarantee zero steady state error when tracking fixed set points in the DC component of the estimated magnetisation flux Φ̂_dc.
As will be readily understood, if a different method is used to estimate the DC component of the estimated magnetisation flux, a different tunable controller may be appropriate. With some such methods a simple proportional control may still be appropriate. It is anticipated that the skilled person will be able to determine a suitable tunable controller without difficulty.
The method uses the DC component of the estimated magnetisation flux to calculate the DC current component that is to be injected into the output transformer to control the flux in the output transformer during a fault situation. This may be done in any manner apparent to the skilled person.
In known methods of current control (i.e., according to the prior art) a reference current is used to calculate a reference voltage that can be used to control the power converter.
Such known methods of current control typically calculate the reference voltage $u_{s}^{*}$
as follows: $u_{s}^{*} = F (i_{s}, i_{s}^{*})$
where i_s is the measured secondary current and $i_{s}^{*}$
is the desired secondary reference current during a short-circuit and that represents a sinusoidal current component.
Such known methods of current control therefore inject current into the output transformer on the basis of a sinusoidal current component only. In the method of the present invention, an additional DC current component may be injected into the output transformer during a short-circuit. In order to do this, the desired secondary reference current during a short-circuit $i_{s}^{*}$
may be modified on the basis of the corrective current which is calculated from the DC component of the estimated magnetisation flux. For example: $i_{s}^{*} = i_{s_ref}^{*} + \frac{1}{K_{F}} {Δi}_{s} = i_{s_ref}^{*} - K_{Φ} {\hat{Φ}}_{dc}$
where K_F is the steady state gain of the current controller, $i_{s_ref}^{*}$
is a secondary reference current that represents the sinusoidal current component and: $- K_{Φ} {\hat{Φ}}_{dc} = - \frac{K_{Φ}^{*}}{K_{F}} {\hat{Φ}}_{dc}$
which corresponds to the desired DC component in the converter current.
The active flux control method is suitable for use with the output transformer connected to any power converter. However, the method is particularly suitable for use with static frequency converters that form part of an electric rail network, as described above. The present invention also provides a controller adapted to control a power converter (e.g., a static frequency converter) to inject a DC current component into an output transformer according to the active flux control method described above.
The present invention also provides an electric circuit including an output transformer connected to a power converter (e.g., a static frequency converter) and a controller, the controller being adapted to:

derive an estimated magnetisation flux of the output transformer using a closed-loop estimator comprising a flux model and a modelled magnetisation flux;
calculate a DC component of the estimated magnetisation flux; and
control the power converter to inject a DC current component into the output transformer, wherein the DC current component is calculated from the DC component of the estimated magnetisation flux.

The controller may include a flux controller (e.g., to derive the estimated magnetisation flux and to calculate the DC component of the estimated magnetisation flux) and a current controller (e.g., to control the power converter to inject the DC current component into the output transformer).
The output transformer and the power converter may form part of a substation for an electric rail network where the output transformer is connected to an overhead line.

Drawings

Figure 1 is a graph showing the voltage behaviour in an electric railway network during a short-circuit without the active flux control of the present invention;
Figure 2 is simplified schematic of a static frequency converter output stage;
Figure 3 is a schematic of the operation of an embodiment of the method of the present invention to control flux in a static frequency converter;
Figure 4 is a block diagram of an active flux controller operating according to an embodiment of the method of the present invention;
Figure 5 is a graph showing the voltage behaviour in an electric railway network during a short-circuit using the active flux control of the present invention; and
Figure 6 is a graph showing the DC component of flux in the output transformer during the events of Figures 1 and 5.

Figure 1 illustrates the voltage oscillation of a 25kV - 50Hz electric railway network during a short-circuit test. The overhead line (or "catenary") of the electric rail network is fed by a static frequency converter (SFC) and the output of the SFC is actively controlled by a current controller according to the prior art that acts to limit the fault current during fault situations.
Initially, during time period A, the network operates with a normal catenary voltage (indicated by solid line 2) and frequency and a zero catenary current (indicated by dashed line 1). At time B a short-circuit occurs, after which, during time period C, the voltage amplitude drops significantly, and the current becomes non-zero and oscillates. At time D, a section of the network is isolated. After which, during time period E, the voltage and current exhibit distortions and oscillations at non-fundamental frequencies. These distortions have negative effects on the operation of protection relays in the electric railway network by introducing multiple zero-crossing events per fundamental cycle. Since protection relays assume sinusoidal waveforms at the fundamental frequency to calculate the exact location of the fault, the distorted waveforms create mis-calculations and wrong isolation of network sections leading to delays in the restoration of the power after a fault. At time F, the fault is cleared.
The appearance of oscillation after the change of impedance during a network fault may be traced to the initial event where the fault occurs. This may be understood with reference to Figure 2, which shows a simplified schematic of an output stage of an SFC for an electric rail network. The output stage comprises a transformer 10 and a filter 11. Although not shown, the SFC and transformer 10 form part of a substation.
The transformer 10 has a primary coil 12 with an inductance L_p and a resistance R_p. The transformer 10 also a secondary coil 13 with an inductance L_s and a resistance R_s. The filter 11 has resistance R_f and a capacitance C_f. The SFC behaves as a voltage source U_sfc and delivers a voltage U_pcc at a point of common coupling 14. The transformer 10 has a magnetisation inductance L_m.
During a short circuit, a voltage at the point of common coupling 14 collapses due to the low impedance of the short circuit. This exposes the transformer 10 to a high voltage variation and the magnetisation flux through the transformer 10 is disrupted. This sudden change generates an aperiodic component (DC component) in the magnetisation flux that may create saturation depending on the instant where the short circuit occurs.
Since the magnetisation flux Φ_m is proportional to the magnetisation current I_m, the flux lags 90° with respect to the voltage at the point of common coupling.
During normal operation, the magnetisation flux Φ_m through the transformer 10 is periodic and its DC component is zero: $u = Ucos (ωt)$
and $Φ_{m} = \frac{U}{ω} \cos (ωt + \frac{π}{2}) + 0_{dc}$
At zero crossing of the transformer terminal voltage, the instantaneous value of the magnetisation flux is at its maximum due to the 90° phase lag. If the short circuit appears at that point, the magnetisation flux is hold close to the maximum, generating an aperiodic component (DC component): $Φ_{m} = \frac{U}{ω} \cos (ωt + \frac{π}{2}) + Φ_{dc}$
Under this situation, the magnetisation flux oscillates around the DC level (i.e., Φ_dc) with a peak value proportional to the network voltage. However, since the voltage on the network is small, due to the low impedance of the short circuit/fault, the maximum flux excursion (AC plus DC components) mildly saturates the output transformer 10 (magnetisation inductance does not change drastically) and no resonances are noticed. This may be seen in Figure 1 during time C where a small catenary voltage appears without distortion.
The situation changes when the impedance during a fault is higher, e.g., a long-distance short circuit or a change of distance to the fault due to the network protection. Due to the increased voltage, the flux, which is already high due to the aperiodic component, goes over the saturation limit. Heavy saturation reduces the magnetisation inductance L_m of the output transformer 10 and generates a different resonance frequency. The effect is periodic since it coincides with maximum values of magnetisation flux (minimum magnetisation inductance) and decays with the aperiodic component of the magnetisation flux. The rate at which the aperiodic component in the flux decays is linked to the value of the network short circuit impedance.
A schematic of the operation of the method of the present invention to control flux in the output transformer is shown in Figure 3. When the method is started (step 1) and in operation, the DC component of the estimated magnetisation flux Φ̂_dc in the output transformer 10 connected to the SFC is monitored (step 2). The SFC and the electric rail network are monitored for network faults (step 3). If no network fault is detected, the method continues to monitor flux (step 2) and no action is taken.
When a network fault is detected, the DC component of the estimated magnetisation flux in the output transformer 10 is compared with an activation (first) threshold value (step 4) and the flux control is activated if the DC component of the estimated magnetisation flux is greater than the activation threshold value (step 5). If the DC component of the estimated magnetisation flux is less than the activation threshold value, the flux control is not activated, and the method simply continues to monitor the DC component (step 2). If the DC component of the estimated magnetisation flux is greater than the activation threshold value, the flux control is activated to inject an appropriate DC current component into the output transformer whilst the DC component of the estimated magnetisation flux in the output transformer continues to be monitored. The flux control continues to inject an appropriate DC current component into the output transformer until the DC component of the estimated magnetisation flux in the output transformer 10 drops below a deactivation (second) threshold value (step 6). The deactivation threshold value is lower than the activation threshold value. Once the DC component of the estimated magnetisation flux drops below the deactivation threshold value, the flux control is deactivated (step 7) and the injection of the DC current component into the output transformer 10 is halted.
In this particular arrangement, the activation threshold value of the DC component of the estimated magnetisation flux in the output transformer 10 is 0.6 p.u. (60% saturation) and the deactivation threshold value of the DC component of the estimated magnetisation flux in the output transformer is 0.5 p.u. (50% saturation). This gives an activation and deactivation function of the active flux control as: $Φ_{{CTRL}_{ON}} = {\begin{matrix} 1, if {\hat{Φ}}_{dc} > 0.6 p . u . \\ 0, if {\hat{Φ}}_{dc} \leq 0.5 p . u . \end{matrix}$
Figure 4 is a block diagram of a controller that can be used to carry out the active flux control when required. The controller includes an active flux controller 20 and a current controller 30.
The active flux controller 20 incudes a flux model 21, a transformer saturation curve 22, a moving average filter 23, and a tunable gain controller 24. The active flux controller 20 has inputs of:

secondary voltage of the output transformer (U_s)
secondary current of the output transformer (I_s)
primary current of the output transformer (I_p)
filter current of the output transformer (I_f)

The output of the active flux controller 20 is a secondary corrective current ΔI_s that represents a DC current component to be applied to the output transformer 10 during the short-circuit fault as described in more detail below. This secondary corrective current ΔI_s is added to a secondary reference current I_{s_ref} that represents a sinusoidal current component to derive a desired secondary reference current $I_{s}^{*}$
during a fault situation. The current controller 30 uses the desired secondary reference current $I_{s}^{*}$
to derive a secondary reference voltage $U_{s}^{*}$
to control the current that is injected into the output transformer 10 by the SFC during the fault situation.
Together the flux model 21 and the transformer saturation curve 22 form a closed-loop estimator for deriving an estimated magnetisation flux Φ̂_m of the output transformer 10.
The flux model 21 is based on an open-loop flux model that estimates the magnetisation flux Φ̂_m from inputs U_s and I_s but with an additional feedback input ΔΦ_m that is derived from the outputs of the flux model 21 and the transformer saturation curve 22 as explained below.
The transformer saturation curve 22 is specifically adapted to the output transformer 10 and has inputs of the measured primary current I_p, secondary current I_s, and filter current I_f of the output transformer 10. These inputs are used to estimate a magnetisation current I_m of the output transformer according to: $I_{m} = I_{s} - I_{p} - I_{f}$
The transformer saturation curve 22 provides a function f_sat for estimating the magnetisation flux Φ_m from the estimated magnetisation current I_m where: $Φ_{m} = f_{sat} (I_{m})$
The estimated flux Φ̂_m from the flux model 21 is subtracted from the estimated magnetisation flux Φ_m derived using the transformer saturation curve 22 and the resulting difference is multiplied by an observer gain 25 (e.g., K_obsv) to produce the feedback input ΔΦ to the flux model 21. The observer gain 25 corrects for potential errors in the estimated magnetisation flux Φ_m that may arise from errors in the model used for estimation.
The flux model 21 calculates an estimate of the magnetisation flux Φ̂_m from the following equations: $U_{s} = L_{s} \frac{{dI}_{s}}{dt} + R_{s} I_{s} + \frac{d {\hat{Φ}}_{m}}{dt}$
where L_s is the secondary inductance of the output transformer 10 and R_s is the secondary resistance of the output transformer.
An observer state is defined as: $x = L_{s} I_{s} + {\hat{Φ}}_{m}$
In an open-loop estimator the observer state equation is: $\frac{dx}{dt} = U_{s} - R_{s} I_{s}$
In the closed-loop estimator, this equation is modified by the feedback input ΔΦ to provide the closed-loop observer state equation: $\frac{dx}{dt} = U_{s} - R_{s} I_{s} + K_{obsv} (Φ_{m} - \hat{Φ})$
This gives an estimate of magnetisation flux Φ̂ by the flux model 21 as: $\hat{Φ} = x - L_{s} I_{s}$
As well as being used to derive the feedback input, the estimated magnetisation flux Φ̂ from the flux model 21 is also input into the moving average filter 23. The moving average filter 23 calculates a DC component of the estimated magnetic flux over N samples: ${\hat{Φ}}_{dc} (t) = \frac{1}{T} \int_{0}^{T} \hat{Φ} (t - τ) dτ$
This equation is cast in a discrete-time recursive form for implementation. The output of the moving average filter 23 is the DC component of the estimated magnetisation flux Φ̂_dc.
A simple proportional controller i.e., a tunable gain controller 24, is used to guarantee zero steady-state error when tracking fixed set points in the DC component of the estimated magnetic flux Φ̂_dc. Hence the secondary corrective current ΔI_s produced by the flux controller 20 is calculated using the tunable gain controller 24 as follows: ${ΔI}_{s} = - K_{Φ}^{*} {\hat{Φ}}_{dc}$
wherein, $K_{Φ}^{*}$
is the gain of the tunable gain controller 24.
The secondary corrective current ΔI_s output by the active flux controller 20 will act to reduce the DC component of the magnetisation flux in the output transformer 10 and is added to the secondary reference current $I_{s_ref}^{*}$
that represents the sinusoidal current component. In particular: $I_{s}^{*} = I_{s_ref}^{*} + \frac{1}{K_{F}} {ΔI}_{s} = I_{s - ref}^{*} - K_{Φ} {\hat{Φ}}_{dc}$
where K_F is the steady state gain of the current controller 30 and: $- K_{Φ} {\hat{Φ}}_{dc} = - \frac{K_{Φ}^{*}}{K_{F}} {\hat{Φ}}_{dc}$
which corresponds to the desired DC component in the converter current.
The secondary reference current $I_{s}^{*}$
is provided as an input to the current controller 30 which derives a secondary reference voltage $U_{s}^{*}$
as follows: $U_{s}^{*} = F (I_{s}, I_{s}^{*})$
where I_s is the measured secondary current in the output transformer 10.
When the active flux controller 20 is operating, the secondary reference voltage $U_{s}^{*}$
will control the SFC to inject current into the output transformer 10 that includes both a DC current component and a sinusoidal current component. When the flux controller 20 is not operating, the secondary reference voltage $U_{s}^{*}$
can be derived using a secondary reference current that does not includes a DC current component but only a sinusoidal current component (e.g., using just $I_{s_ref}^{*}$
).
As set out above, the active flux control method of the present invention acts to inject a current into the output transformer 10 connected to an SFC of an electric rail network to control an aperiodic component of the magnetisation flux during a fault situation. The simulated application of the active flux control method is shown in Figure 5.
Figure 5 illustrates the voltage oscillation of a 25kV - 50Hz electric railway network during the same short-circuit test illustrated in Figure 1, with the addition of the active flux method of the present invention described above. That is, the overhead line of the electric railway network is fed by the SFC and the output of the SFC is actively controlled by the current controller 30 in combination with the flux controller 20 during a fault situation.
Initially, during time period A, the network operates with a normal catenary voltage (indicated by solid line 2) and frequency and a zero catenary current (indicated by dashed line 1). At time B, a short-circuit occurs, after which, during time period C, the voltage amplitude drops significantly, and the current becomes non-zero and oscillates. At time D, the network section is isolated. After which, during time period E, the aperiodic component of the magnetisation flux is controlled by the method of the present invention to reduce the distortions and oscillations at non-fundamental frequencies shown in Figure 1. At time F, the flux control of the method of the present invention is turned off.
Figure 6 shows a comparison of the aperiodic (DC) component of the magnetisation flux in the output transformer 10 during the events shown in Figures 1 and 5. The DC component of the magnetisation flux of Figure 1 is indicated by solid line 40 and the DC component of the magnetisation flux of Figure 5 is indicated by dotted line 41. This shows that actively injecting a DC current component into the output transformer 10 acts to reduce the saturation level in the output transformer.
Figure 6 also shows the activation threshold 43 and deactivation threshold 44 for the active flux control of the present invention. The active flux control is activated after time B when the magnitude of the DC component of the magnetisation flux in the output transformer 10 exceeds the activation threshold 43. The active flux control remains activated until the magnitude of the DC component of the magnetisation flux in the output transformer 10 drops to the deactivation threshold 44. The activation threshold 43 is 0.6 p.u. and the deactivation threshold 44 is 0.5 p.u.
It is to be understood that the above description of a specific arrangement of the invention is illustrative only and is not intended to be limiting. Unless otherwise indicated by context any individual feature of the specific arrangement may be included in an arrangement of the invention independently of any other feature.

Claims

A method of active flux control for an output transformer (10) connected to a power converter comprising the steps of:
deriving an estimated magnetisation flux of the output transformer (10) using a closed-loop estimator comprising a flux model (21) and a modelled magnetisation flux (22);

calculating a direct current DC component of the estimated magnetisation flux; and

carrying out current control by injecting a DC current component into the output transformer (10), wherein the DC current component is calculated from the DC component of the estimated magnetisation flux.
A method according to claim 1, wherein the modelled magnetisation flux (22) is obtained from a magnetisation curve adapted to the output transformer (10) and an estimate of a magnetisation current of the output transformer (10).
A method according to claim 2, wherein the estimate of the magnetisation current is calculated from a measured primary current, a secondary current, and a filter current.
A method according to any preceding claim, wherein the DC component of the estimated magnetisation flux is estimated using a moving average filter (23) over N samples.
A method according to any preceding claim, wherein the flux model (21) inputs are:
a measured secondary current;

a measured converter voltage; and

an input derived from the difference between an output of the flux model (21) and an output of the modelled magnetisation curve (22).
A method according to any preceding claim, further comprising using an activation function that activates the current control when the DC component of the estimated magnetisation flux is above a first threshold value and subsequently deactivates the current control when the DC component of the estimated magnetisation flux is equal to or below a second threshold value, wherein the second threshold value is lower than the first threshold value.
A method according to claim 6, wherein the first threshold value is between 20% saturation and 90% saturation of the output transformer (10).
A method according to claim 6 or claim 7, wherein the second threshold value is between 10% saturation and 80% saturation of the output transformer (10).
A method according to any of claims 6 to 8, wherein the first threshold value is greater than the second threshold value by at least 5% of the saturation of the output transformer (10).
A method according to claim 6, wherein the first threshold value is 50% saturation of the output transformer (10) and the second threshold value is 40% saturation of the output transformer; or
wherein the first threshold value is 60% saturation of the output transformer and the second threshold value if 50% saturation of the output transformer.
A method according to any preceding claim, further comprising using a tunable controller (24) to control the magnitude and/or duration of the DC current component, said tunable controller (24) acting on the DC component of the estimated magnetisation flux.
A method according to any preceding claim, wherein the current control includes:
calculating a corrective current from the DC component of the estimated magnetisation flux;

providing a reference current; and

calculating the total current to be injected into the output transformer (10) from the corrective current and the reference current.
A controller (20, 30) adapted to control a power converter to inject a DC current component into an output transformer (10) according to the method of any preceding claim.
An electric circuit including an output transformer (10) connected to a power converter and a controller (20, 30), the controller (20, 30) being adapted to:
derive an estimated magnetisation flux of the output transformer (10) using a closed-loop estimator comprising a flux model (21) and a modelled magnetisation flux (22);

calculate a DC component of the estimated magnetisation flux; and

control the power converter to inject a DC current component into the output transformer, wherein the DC current component is calculated from the DC component of the estimated magnetisation flux.
An electric rail network comprising the electric circuit according to claim 14 and an overhead line connected to the output transformer.